From the monthly archives:

May 2008

Play tag at Artigo

by Denise on May 22, 2008

If you're new here, you may want to subscribe to my RSS feed. Thanks for visiting!

How it works

Artigo is another site that, like Labelme, Peekaboo, and steve.museum, is geared toward using the collective intelligence of internet users to gather data that will improve image search functionality.  Artigo is most like steve.museum in being semantically oriented.  As at steve.museum, you provide your own tags for the image that is presented.  Artigo’s image database contains 15,000 images and is set up at the University of Munich.  But there are several refinements that make Artigo game-like, and arguably more effective in obtaining valid results.

Artigo - Mid Game

Artigo - Mid Game

Each game session is timed; the players have 5 minutes. The number of pictures reviewed is a function of how quickly the players agree to move on to the next one.  Generally you see 5-7 images per game.  There are TABOO words, ones that have already been associated with the image.  These appear below the image in RED.  Players must dig deeper to find appropriate tags.

The more taboo words there are for an image the more points you earn for the tags you provide.  Most importantly your tag is accepted only if you and your partner have both provided it, thus weeding out the chaff.

As you play a countdown of the seconds remaining appears at the top, and your points up to that point on the bottom right.  You can see how many words your partner has selected, they appear as BLUE dots on the left, but not what they are.  But when the game is over, each image is shown again with artist, title and date identified.

You also see the tags given by your partner and yourself, and any that agree are shown in  YELLOW.   They are the ones that you get credit for.  Of course you get your final score as well.

Artigo also offers a monetary inducement to play.    [click to continue…]

  • Share/Bookmark

{ 0 comments }

CLICK!

by Denise on May 17, 2008

CLICK! submission     Photo credit: Elyse Taylor

CLICK! submission Photo credit: Elyse Taylor

A current exhibit at the Brooklyn Museum of Art (BMA) called CLICK! is a great example of the creative role that museums can play in utilizing the web to involve users.  Click! is a multi stage project.  First, there was an open call for photographs for the exhibit on the theme of the “Changing Faces of Brooklyn.”  Then an open call for evaluation of the submitted photographs online was issued.  Finally an exhibit at the museum of the top 20% of the 389 photos submitted was assembled.  But that isn’t all.

There is also an exhibit of all the photographs, and multiple statistical views of the voting.  These results include very interesting data about the relative ratings of the raters (self-identified in 5 categories from naive to expert), along with a host of other data. I found this information as interesting as the exhibit itself.  Begging won’t help.  You’ll just have to visit the site yourself to find out how the ratings of experts compared to the ratings of the non-experts.  The site has lots of other great features, including an intelligent commentary, a look into the process of mounting the physical exhibit, the comments of raters about the photos, a blog, a podcast, and, novel to me, [click to continue…]

  • Share/Bookmark

{ 0 comments }

Peekaboom

by Denise on May 12, 2008

Peekaboom is a blast.  It’s a totally addictive online game that also contributes raw data to CBIR researchers.

Peekaboom Screenshot

“We introduce Peekaboom, an entertaining web-based game that can help computers locate objects in images. People play the game because of its entertainment value, and as a side effect of them playing, we collect valuable image metadata, such as which pixels belong to which object in the image. The collected data could be applied towards constructing more accurate computer vision algorithms, which require massive amounts of training and testing data not currently available.

Peekaboom has been played by thousands of people, some of whom have spent over 12 hours a day playing, and thus far has generated millions of data points. In addition to its purely utilitarian aspect, Peekaboom is an example of a new, emerging class of games, which not only bring people together for leisure purposes, but also exist to improve artificial intelligence. Such games appeal to a general audience, while providing answers to problems that computers cannot yet solve.” Peekaboom: A Game for Locating Objects in Images

News: Visionary Research: Teaching Computers to See Like a Human

Reblog this post [with Zemanta]
  • Share/Bookmark

{ 0 comments }

Ideal Access to Images

by Denise on May 6, 2008

Eden as depicted in the first or left panel of...

Eden as depicted in the first or left panel of The Garden of Earthly Delights Hieronymus Bosch Image via Wikipedia

The potential usefulness of well cataloged images seems enormous.  Images can open up avenues of thought and research previously unavailable, not just within the narrow category that the image may be typically associated with (whether that be art history, architecture, geology, or some other area) but across disciplines, and into wholly new areas.  While cross disciplinary studies are not new, ready access to the whole store of images would greatly facilitate and stimulate further work by artists, economists, historians, and doctors, among others, independently and collaboratively.  Leaving aside the myriad barriers posed by expense, expertise, technology, and legal considerations (including copyright), and the differing views about priorities and cost allocation, here is my attempt to articulate an ideal.

Universal, scalable, trustworthy, sustainable, and permanent access to all heritage digital images from the publicly accessible to the privately held, including those from all museums and cultural institutions, archives and libraries, private collections, and publishers, using a system of organization that maximizes entry points and interconnections between objects and relevant texts, to facilitate retrieval for all imaginable purposes, through interfaces which have the capability of searching across all platforms, providing high resolution images, thumbnail browsing, copying and reuse for non-commercial purposes, and links to contextual materials, with continuously updated information of the applicability of worldwide copyright status and contact information for obtaining permissions.

This ideal is more utopian than anything; this is what brought Hieronymus Bosch’s The Garden of Earthly Delights to mind as a good illustration here.   My utopian ideal does provides a context for looking at what exists, and what is evolving.  If there could be agreement about the goal, however unrealistic at present, this could inform the current and developing practices at critical institutions.

Since first writing this piece, The Garden of Earthly Delights has become a prime example of a different aspect of ideal access to images.  Google Earth chose to render this painting, along with  13 others at the Prado,  in unimaginably high resolution.  The detail below is of the face of the monkey that is  sitting atop an elephant in the distance, about a third of the way from the top of the picture.   In the thumbnail of the painting here, the monkey is barely detectable.

Garden of Earthly Delights - detail of moneky face

Garden of Earthly Delights - detail of moneky face

To view this painting in this way is truly breathtaking.   Google’s accomplishment far exceeds my imagination; maybe the rest of the utopian dram described above is also possible.

I must  unfortunately note the Museo del Prado’s non-existent subject search capability.   While the collections page does say that search by subject is possible: “Access on line to approximately 2000 works of the Museum’s collection. …This data base will enlarge until it holds the complete collection….The advanced search engine facilitates consultation, using categories such as artist, title of work, subject, chronology and reference number…” in fact it is not.   The advanced search has drop down boxes with NO options, and “subject” is not even one of them.   On top of that, the word “Chronology” is misspelled.  The Prado’s aspiration to digitize and share the entire collection is admirable, but image cataloging and search functionality should go hand and hand with the digitization.  It is well known that once the digitization process is complete it is very rare for more metadata to be added at a later time.

Reblog this post [with Zemanta]
  • Share/Bookmark

{ 0 comments }